ABSTRACT
This chapter overviews a software rejuvenation scheduling problem and provides a comprehensive survey of the optimal periodic software rejuvenation policies in discrete time under four criteria of optimality: expected cost per unit time in the steady state, steady-state system availability, cost-effectiveness, and expected total discounted cost over an infinite time horizon. Present-day applications in computer systems impose stringent requirements in terms of software dependability, because system failure, caused by software failure in almost all cases, may lead to a huge economic loss or risk to human life. The system operation of the software system starts at time n = 0 in the highly robust state. For some reason, such that the total amount of memory leaking attains a critical threshold, the process makes a transition to the failure probable state after the time period Z elapses. Cost-effectiveness is defined by combining the expected cost per unit time in the steady state and the steady-state system availability.
